For 2020, your company is thinking “We are on our way to level up our business. Competition is incredibly fierce and we need something practical to get us there.” This article covers practical ways businesses use cutting-edge artificial intelligence (AI), machine learning (ML), and deep learning (DL) that you could achieve especially combined with best of breed API Management including Axway’s AMPLIFY platform. I discussed with Emmelyn Wang, Axway’s Digital and API Business Strategist on the Catalyst Team, why the common link to intelligent models for faster, more accurate business decisions is how you harness your data. As Emmelyn was writing the Disruptive 2020 API Trends article, the feedback she received was skepticism about AI/ML which is common. So, we wanted to dive in deeper to talk about production-grade capabilities out in the world today.
Adaptive Business Intelligence and Practical Applications of AI
We’re far from achieving the strong AI depicted in movies like Minority Report or The Island. However, the AI in our daily lives comes in the form of digital assistants by way of Siri, Cortana or Alexa. The subset of AI and ML, called deep learning is able to process complex tasks that include recognizing image and sound using multi-layered networks leads the charge to artificial general intelligence or strong AI.
How does AI/ML/DL relate to API management and hybrid integration platforms? More importantly, how do they all relate to faster decision making, which makes even the largest enterprises more nimble and innovative?
Over the past eight months at the time of writing, over 1400 IT professionals have taken Axway’s digital transformation maturity assessment. The overall trend we discovered is that only one third of respondents have an active plan to compete and fend off disruption with solutions like intelligence.
one third of respondents have an active plan to compete and fend off disruption with solutions like intelligence.
Lofty goals of using the company’s various data sources to achieve business intelligence based on models of reuse seem out of the question. Company leaders keep asking themselves “How can we work smarter and be less reactive?”
- What’s really possible for machine learning as the brain to provide recommendations to decision-makers both inside and outside of the company?
- How do you use API-driven intelligence to feed the data model?
- How do you manage data flows that serve as inputs and outputs of APIs and Algorithms?
With Axway’s AMPLIFY stack, you can build predictions based on machine learning. The API that you publish and/or consume using Axway’s APIM becomes the input for predictive algorithms and pattern recognition using the AMPLIFY platform. Once data is designed to be available for processing, your company no longer has to have specialized artificial intelligence AI/ML/DL knowledge or expertise to translate information so that business leaders can understand how to make insightful predictions.
I’ve spoken at several conferences globally about the role of streaming APIs in the financial sector, including my research on high frequency trading HFT. In my talks, I hone in on key algorithms and their business applications.
AI and machine learning
Let’s look at this subset of ML-based streams of data. The standard approach is batch-based, finite training sets and static models. The data stream approach means infinite training sets and approaches ML with dynamic models. The data model updates itself and autocorrects for every decision being made. For example, the Massive Online Analysis set of algorithms performs machine learning at scale for scenarios that process concept drift and big data streams in real time. The algorithms are accessible via GUI, command line, and the Java API.
You can apply ML to
- Stock market and other economic predictions
- Delivery route planning and optimization
- Drug discovery and development
- Natural language processing including translation and language generation
- Optimizing any kind of buying experience from eCommerce to real estate
It’s also helpful to know that
- Fuzzy logic is closer to how the human brain works and helps with natural language processing
- Genetic programming is a computational model that processes complex problems by continually approaching the solution through testing and selecting the best choice based on what is most effective or functional.
- Bayesian logic processes historical events to predict how future events will occur
Emmelyn asked a few experienced mobile app developers whose businesses rely on delivering mobile capabilities — how easy is it to integrate AI into your app today with available tools? Most app developers shrug since it is difficult to officially quantify. Back when the craze of “organic” foods hit grocery store shelves worldwide, quality groups needed to determine the criteria of the designation. Today, when a website, app, or business service claims to be powered by AI, businesses and consumers become skeptical. How do you prove that the engine is really using AI when it is really only using a basic decision tree? Can an experience be “certified” as being powered by AI?
The common way to prove that AI and APIs work well together is large sets of data, models, and the algorithms that transform the data between the two to train machines to more quickly predict what it takes humans much longer to do.
The common way to prove that AI and APIs work well together is large sets of data, models, and the algorithms that transform the data between the two to train machines to more quickly predict what it takes humans much longer to do. Training machines to help business users process more data accurately means we can do a lot more in strategic and high-level direction setting.
The magic 8 ball is a nice physical depiction of a data model that is famous for answer-seeking consumers. However, on a macro scale, hedge funds use data models from a shipping API to predict the volume of goods being shipped around the world to understand real-time economic growth. And from a security perspective, machine learning helps institutions and companies actively prevent fraud. At home, we have applications like Netflix, Hulu, Disney+ and other programming recommendations on our favorite digital media apps across devices. The front end API provides the recommendations consumed by any device: mobile, tablet, TV, and computer.
Here are more practical ways we enable our customers and examples of production-grade AI/ML/DL and APIM capabilities that power decisions today:
- Transport and Logistics – Optimal route planning and delivery information powers more than just resource management | Case Study – Distributors which include global retailers, eCommerce companies, and rideshare applications work to make sure all major players on their platforms efficiently operate whether you’re predicting loads for third party logistics or you have your own fleet to manage. In general, resource and route planning for businesses and consumers works better when you power it with ML.
- Food and Agriculture – An active prevention method saves companies millions of dollars upstream and downstream. John Deere is famous for proactively visiting farmers by detecting exactly which part in the tractor needs maintenance or replacement. | Case Study – A few of our customers in this industry use IoT sensors and other field data collection devices combined with hyperledger technology to keep livestock healthy.
- Financial Services – Identity protection, prevent business fraud, investment decisions | Case Study – Credit card protection and company and consumer identity verification are all ways to actively prevent fraud.
- Oil and Gas – Safety, resource allocation, save the environment | Case Study – We see various companies in the oil and gas ecosystem gathering information from rigs to refineries and capture data from hardware and field engineers to more operate in optimal conditions. Since mining the earth’s resources requires specialized knowledge from satellites and geologists and petroleum engineers, the sheer amount of data gathered can be used to power many other industries besides its own ecosystem. The data can also be used to keep the company compliant and safe which is important for the workforce and its leaders. These companies can detect issues before an oil spill ever happens and aid in faster, efficient clean ups if a spill does occur.
- Managing Digital Business Experiences – How can start ups, SMBs and Enterprises deliver better Customer Expectations | Case Studies
- Digital Lead Nurturing and Support Streams
The idea is to use best of breed services and tooling so that you don’t need to have in-house AI/ML/DL expertise to leverage data and APIs. You can avoid vendor lock in by delivering value quickly and future proof how you scale.
Call for Discussion
What are ways that your organization or business unit uses APIs to power AI in your products or driving innovation today?
Read Emmelyn’s article “12 Disruptive API Trends for 2020.”